Improving the analysis of global value chains: the UNCTAD ...

115

Improving the analysis of global value chains: the UNCTAD-Eora Database

Bruno Casella, Richard Bolwijn, Daniel Moran and Keiichiro Kanemoto*

The UNCTAD-Eora Global Value Chain (GVC) database offers global coverage (189 countries and a "Rest of World" region) and a timeseries from 1990 to 2018, reporting on key GVC indicators. This paper explains the methodology for compiling the UNCTAD-Eora GVC database, including nowcasting employed in the estimation of recent years; second, it provides a comparison of the results against other value-added trade databases, with a focus on the OECD Trade in Value Added (TiVA) dataset; and lastly discusses the relevance of GVC data for the analysis of globalisation patterns, particularly at the intersection between trade, investment and development. Keywords: trade in value added; MRIO; global value chains; complex value chains; value added in export; input-output analysis

1. Introduction

A pivotal element in the analysis of international production are global value chains (GVCs), which are fragmented and geographically dispersed production processes where different stages are located across different countries. GVCs are coordinated by multinational enterprises (MNEs) investing in productive assets worldwide and trading inputs and outputs intra-firm, at arm's length or through their network of non-equity mode (NEM) partners. UNCTAD estimates that up to 80 per cent of global trade involves MNEs (World Investment Report 2013). In this respect, the analysis of GVCs is fully complementary to the analysis of FDI and international production.

* Bruno Casella and Richard Bolwijn are at the United Conference on Trade and Development. Daniel Moran and Keiichiro Kanemoto work at Eora. Correspondence with the authors may be addressed jointly to Bruno Casella (Bruno.Casella@) and the Eora MRIO maintainers (info@worldmrio. com). The views expressed in this paper are solely those of the authors.

116

TRANSNATIONAL CORPORATIONS Volume 26, 2019, Number 3

Recently, major analytical developments in the treatment of inter-country inputoutput tables have opened new avenues for the empirical research on global value chains. In particular, the availability of databases that break down trade according to the origin of its value added ("value added trade" or "value added in exports" data) enables the analysis of GVC patterns by countries and industries, at a level of granularity that was unimaginable as recent as ten years ago. The most important cross-regional value-added trade databases include the UNCTAD-Eora GVC database, the World Input-Output Database (WIOD) and the OECD's Trade in Value Added Database (TiVA). Major regional initiatives include the Asian Multi-Region Input-Output Database from the Asian Development Bank and the South-American Input-Output Table from the Economic Commission for Latin America and the Caribbean (ECLAC). Table 1 provides an account and a comparison of the different and ongoing initiatives to map GVCs (see also Tukker and Dietzenbacker, 2013).

The UNCTAD-Eora GVC database was initially launched in the context of the analysis conducted for the World Investment Report 2013 (WIR13), with its main theme "Global Value Chains: Investment and Trade for Development" (UNCTAD, 2013). Compared with alternative databases, its distinctive feature is broad geographical coverage, including virtually all countries. Owing to this comprehensive coverage the database has become the preferred reference source of value-added trade data in analysis involving developing economies (AfDB, OECD, & UNDP 2014; UNECA, 2015; UNIDO, 2016; IMF, 2015a; IMF 2015b; IMF 2016a; IMF 2016b).

Given the importance of GVC analysis in the context of globalization and development and the high demand for value-added trade data, particularly for developing countries, UNCTAD-Eora has upgraded its GVC database. This has led not only to an update of the 2013 dataset to include GVC indicators up to 2015 but also a new improved version, featuring a "nowcast" methodology to project value-added trade data from 2016 to 2018. This step addresses one of the main weaknesses of available value-added trade databases (including the WIOD, TiVA and the previous version of the UNCTAD-Eora GVC database), namely the time lag of two to three years between the most recent data and the time of the analysis. A further update of the UNCTAD-Eora GVC database, including GVC indicators for 2016 and 2017 based on actual data, is in preparation and will be published in conjunction with this paper.

The main outcome of the UNCTAD-Eora database is a set of basic GVC indicators, including foreign value added (foreign value embedded in a country's exports), domestic value added (domestic value embedded in a country's exports) and domestic value added embedded in other countries' exports. Other important GVC indicators, such as GVC participation, can be easily computed from the three basic indicators (Koopman et al., 2014). UNCTAD-Eora GVC indicators are

Improving the analysis of global value chains: the UNCTAD-Eora Database

Table 1. Efforts to map GVCs (status as of August 2019)

Project

Institution

Data sources

Countries Industries

Years

Comments

UNCTAD-Eora GVC Database

Trade in Value Added (TiVA) dataset

World Input-Output Database (WIOD): 2016 Release

EXIOBASE

ADB Multi-Region Input-Output Database (ADB MRIO)

Global Trade Analysis Project (GTAP)

South American Input-Output table

Source: UNCTAD.

UNCTAD/Eora OECD

National Supply-Use and I-O tables, 189 and I-O tables from Eurostat, IDEJETRO and OECD

National I-O tables

64

26-500 depending on the country

34

1990?2015 (nowcast for 2016, 2017 and 2018)

Meta database drawing together many sources and interpolating missing points to provide broad, consistent coverage

2005?2015

Information on all OECD countries, and 27 non-

(projections 2016) member economies (including all G20 countries)

Consortium of 11

National Supply-Use tables

43

56

institutions, EU funded

2000?2014

Other multi-region input-output databases

EU-based consortium, National supply-use tables exiobase.eu

44+5

200

1995?2013

Asian Development

An extension of WIOD which

45

35

Bank

includes 5 additional Asian

economies (Bangladesh, Malaysia,

Philippines, Thailand and Viet Nam)

2000, 2005? 2008, 2011

Purdue University

Contributions from individual researchers and organizations

121

65

countries

plus 20

regions

2004, 2007, 2011, 2014

ECLAC and Institute of Applied Economic Research (IPEA) from Brazil

National I-O tables

10

40

2005

Based on of cial national account statistics; uses end-use classi cation to allocate ows across partners and countries

Covers 44 countries plus ve rest-of-world regions

The information for the 5 additional Asian countries are estimates methodically produced to assist research and analysis, not of cial statistics

Includes data on areas such as energy volumes, land use, carbon dioxide emissions and international migration.

Based on of cial information from National Accounts

117

118

TRANSNATIONAL CORPORATIONS Volume 26, 2019, Number 3

publicly available at granular year-, country- and industry-level on the UNCTADEora webpage.1

The intention is to establish the UNCTAD-Eora project as a continuing project for the update and improvement of GVC data and analysis, with annual updates envisaged.

In this context, this paper has two objectives: First, it presents the analytic and methodological construction of the UNCTAD-Eora database (sections 2 and 3). Second, it compares results with other available databases, particularly the OECD TiVA, for data validation purposes (section 4). The concluding section puts the UNCTAD-Eora database in the broad context of the analysis of the tradeinvestment-development nexus: it shows how GVC data can provide an important perspective on some relevant trends at the intersection between these three key areas in modern globalization.

2. The analytical background of the new UNCTAD-Eora database

In this section we briefly retrace the steps that lead to the establishment of the new UNCTAD-Eora database. The first step (section 2.1) ? the construction of a multiregional input-output (MRIO) dataset ? is the most technically complex and computationally intensive. We present it only qualitatively; for more detail the existing literature is referenced. Once an MRIO is available, some straightforward algebraic steps allow to fit the relevant information contained in the MRIO into the framework of value-added trade and derive the key GVC indicators (section 2.2). Finally, a nowcasting procedure is implemented to project value-added trade data from the last available year onward (section 2.3). Unlike section 2.1 and section 2.2 which are essentially summaries of existing material, the treatment of nowcasting in section 2.3 is new, hence its analytical elaboration here is more detailed.

2.1. The construction of the Eora MRIO dataset

This section provides an overview of how the Eora MRIO is constructed. For a more comprehensive explanation, the primary reference paper is Lenzen et al. (2012). Some more approachable summary papers are Lenzen et al. (2013); Moran and

1 . For references to the UNCTAD-Eora database, cite this method paper as follows: Casella, B. et al. (2019). Improving the analysis of global value chains: the UNCTAD-Eora Database, Transnational Corporations Journal 26(3). New York and Geneva: United Nations.

Improving the analysis of global value chains: the UNCTAD-Eora Database

119

Geschke (2013); and Moran (2013). The documentation section of the Eora website (at ) also provides several papers and reports that present the main elements of I/O analysis.

The Eora dataset provides a multi-region input-output table at the global level to estimate value added in trade. The construction of the Eora MRIO table follows several steps.

a. The starting points are the national IO tables or supply/use tables (SUTs). National SUTs are recommended over input-output tables because they provide information on both products and industries. However, the national statistics bureaus in some countries still provide only input-output tables. A supply table provides information on products produced by each domestic industry and a "use" table indicates the use of products by industries or final users. As SUTs are only available for a limited number of countries, the remaining countries are hence represented by input-output (I/O) tables, which can be sourced from available data or compiled according to a range of assumptions. In order to avoid departures from the original raw data, EORA preserves the sectoral classification from each data provider. The complete list of raw data sources involved in preparing the IO table for each country in Eora is available at the Quality Report section of the Eora website and in the Supplementary Information of Lenzen et al. (2012).

b. National SUTs and I/O tables are linked through international trade statistics using import tables to obtain a multi-region input-output table. At this step, an estimation procedure is used to construct so-called "off-diagonal" trade blocks, estimating flows from each export sector in each origin country (rows) to each importing sector in each destination country (columns). Trade data is most often reported by product and by producer and consumer country. However, an offdiagonal trade block in an IO table requires knowing how goods from each exporting sector are absorbed into each importing sector. Put another way, the raw data is three-dimensional, but the IO table requires four dimensions. Thus, creating the trade blocks involves several assumptions and estimation steps. The challenges and procedures used to estimate trade are presented in full in Lenzen et al. (2012).

c. After obtaining a first estimate of an MRIO table, the resulting trade data are balanced through an industry-level balancing condition: the total output produced by each sector must equal the sum of the inputs used by that sector. This has been achieved via "constraints data": i) Input-output tables and main aggregates data from national statistics offices; ii) Input-output compendia from Eurostat, IDE-JETRO and OECD; iii) The UN National Accounts Main Aggregates Database and official country data; iv) The UN COMTRADE and UN Service Trade international trade databases. An optimization procedure (a variant of the

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download